fix: do not use excessive memory when rendering markdown (#382)

Due to legacy reasons we enabled the emoji plugin from pymdown.
Every time the Markdown object gets created, it will load the
emoji extension and will deep copy the emoji database.
This is about 150kb of data:
len(str(pymdownx.emoji1_db.emoji))

This is not a lot, but it is unnecessary and can be avoided.

solara/components/markdown.py

@@ -7,6 +7,7 @@
 from typing import Any, Dict, List, Union
 
 import ipyvuetify as v
+import pymdownx.emoji
 import pymdownx.highlight
 import pymdownx.superfences
 
@@ -183,6 +184,13 @@ def highlight_code(code, name, attrs):
     return v.VuetifyTemplate.element(template=_markdown_template(html)).key(hash)
 
 
+_index = pymdownx.emoji.emojione(None, None)
+
+
+def _no_deep_copy_emojione(options, md):
+    return _index
+
+
 @solara.component
 def Markdown(md_text: str, unsafe_solara_execute=False, style: Union[str, Dict, None] = None):
     """Renders markdown text
@@ -248,6 +256,9 @@ def highlight(src, language, *args, **kwargs):
                 "tables",
             ],
             extension_configs={
+                "pymdownx.emoji": {
+                    "emoji_index": _no_deep_copy_emojione,
+                },
                 "pymdownx.superfences": {
                     "custom_fences": [
                         {